Peter Lietz

Thank you very much for the quick answer!

Just in case anybody else stumbles upon this, the workaround I am now using is to convert to "str". Not ideal, but does the trick. ``` df = pl.DataFrame({"a": [1,...

I can confirm this. BTW, after changing the model sense to minimize and negating the objective, the solution objectives are correctly reported as -13 and -12. I suspect the bug...

PS: codon is in the lead slightly on fedora 38. However, I can produce a variant of the code where codon performance is worse than python's on fedora as well....

Thank you very much, @marioroy, that is an excellent idea. After memoization, the codon version is 31% faster than python.